"We were single-threaded on one vendor, one AI vendor at the Department of War," Under Secretary of Defense for Research and Engineering Emil Michael told an audience Thursday at the Special Competitive Studies Project's AI+ Expo. "But never again we'll be single-threaded with any one model," he added, framing a rapid, deliberate pivot inside the Pentagon away from a sole frontier-AI supplier.
Emil Michael's public pledge to diversify frontier AI suppliers
Michael made the remarks as part of a broader defense effort to break dependence on a single provider for advanced artificial intelligence systems. He described the previous posture as one where "these are sophisticated, protective systems that take a lot of work to integrate on, so it wasn't like I could just turn on a few other models that easily." His vow — that the Department of Defense "will no longer rely on a single artificial intelligence provider" — reaffirms a deliberate shift toward multiple vendors for AI integrated into defense and intelligence uses.
Pentagon agreements with multiple AI companies for classified deployments
The Pentagon recently announced agreements with multiple AI companies to expand frontier AI deployments across classified environments. The source describes those agreements as "widely viewed as an effort to reduce reliance on Anthropic while maintaining access to advanced AI capabilities for national security missions." Officials told the outlet these moves are intended to keep access to advanced capabilities used in cyber defense, intelligence analysis and military operations while diversifying supplier risk.
Anthropic dispute: supply-chain designation and ongoing litigation
The administration's dispute with Anthropic sits at the center of the policy change. Earlier this year the Defense Department designated Anthropic a supply-chain risk and moved to restrict its systems from defense environments, prompting an ongoing legal battle between the company and the government. The source notes the legal fight has escalated, with court rulings referenced in coverage under the headline "Court Backs Pentagon Anthropic Ban - But the Fight Continues."
Anthropic has pushed back in court filings, arguing that once its systems are deployed inside secure government environments the company lacks any technical ability "to remotely alter, disable or interfere with them." The dispute has become an early test of whether federal agencies can simultaneously reduce dependence on a major AI vendor and retain access to the frontier models officials say are increasingly critical to national-security missions.
White House guidance and the prospect of pre-release government review
Michael's comments echo broader White House initiatives described in the reporting. New reporting shows the White House is preparing updated national security guidance expected to direct agencies to diversify AI vendors while establishing clearer rules around how advanced systems are deployed across military and intelligence environments. The White House is also reportedly considering whether frontier AI systems should undergo formal government review before deployment or release, amid concern that models capable of improving cyber defense could also speed vulnerability discovery, exploit development and offensive cyber operations.
What this means for the Pentagon, Anthropic, and major AI developers
- Pentagon and defense/intelligence operations: The department will expand contracts and testing arrangements with multiple vendors to operate frontier AI inside classified environments, aiming to reduce single-vendor supply-chain risk while preserving operational access to advanced capabilities.
- Anthropic: The company remains the subject of a supply-chain risk designation and an ongoing legal dispute; in court filings Anthropic contends it cannot remotely alter systems once deployed in secure government environments.
- Major AI developers: Other providers now face accelerated government engagement for classified deployments and potential new testing regimes; they will be part of the Pentagon's immediate diversification strategy and any future pre-release or formal review processes the White House endorses.
The record in the reporting ties three simultaneous threads together: a Pentagon decision to stop relying on a single frontier-AI supplier, public agreements to broaden classified deployments with multiple companies, and an active legal confrontation with Anthropic that helped spur that shift. With the White House preparing guidance that would institutionalize vendor diversification and possibly require formal review of advanced models, the next steps are likely to be administrative and legal — expanding vendor pipelines, negotiating terms for classified integrations, and resolving the court fight that has already altered procurement practices.
https://www.govinfosecurity.com/pentagon-official-vows-to-diversify-frontier-ai-suppliers-a-31632
